scsi: scsi_debug: Fix do_device_access() handling of unexpected SG copy length
PATCH 1/1 scsi: scsi_debug: Fix do_device_access() handling of unexpected SG copy length
If the sg_copy_buffer() call returns less than sdebug_sector_size, then we drop out of the copy loop. However, we still report that we copied the full expected amount, which is not proper.
Fix by keeping a running total and return that value.
drivers/scsi/scsi_debug.c +4 -6
@@ -3651,7 +3651,7 @@ static int do_device_access(struct sdeb_store_info *sip, struct scsi_cmnd *scp,enum dma_data_direction dir;struct scsi_data_buffer *sdb = &scp->sdb;u8 *fsp; -int i; +int i, total =0 ;/* * Even though reads are inherently atomic (in this driver), we expect
@@ -3688,18 +3688,16 @@ static int do_device_access(struct sdeb_store_info *sip, struct scsi_cmnd *scp, fsp + (block * sdebug_sector_size), sdebug_sector_size, sg_skip, do_write);sdeb_data_sector_unlock (sip, do_write); -if (ret != sdebug_sector_size) { - ret += (i * sdebug_sector_size); + total += ret; +if (ret != sdebug_sector_size)break ; - } sg_skip += sdebug_sector_size;if (++block >= sdebug_store_sectors) block =0 ; } - ret = num * sdebug_sector_size;sdeb_data_unlock (sip, atomic); -return ret; +return total; }/* Returns number of bytes copied or -1 if error. */
Signature · 1 line
2.31.1
Quoted 10 lines · John Garry
If the
sg_copy_buffer()call returns less thansdebug_sector_size, then we drop out of the copy loop. However, we still report that we copied the full expected amount, which is not proper.Fix by keeping a running total and return that value.
Fixes 84f3a3c01d70 ("scsi: scsi_debug: Atomic write support")Reported-by Colin Ian King <[email protected]>Suggested-by Dan Carpenter <[email protected]>Signed-off-by John Garry <[email protected]>
Looks good to me.
regards,
dan carpenter
Quoted 36 lines · John Garry
If the
sg_copy_buffer()call returns less thansdebug_sector_size, then we drop out of the copy loop. However, we still report that we copied the full expected amount, which is not proper.Fix by keeping a running total and return that value.
Fixes 84f3a3c01d70 ("scsi: scsi_debug: Atomic write support")Reported-by Colin Ian King <[email protected]>Suggested-by Dan Carpenter <[email protected]>Signed-off-by John Garry <[email protected]>drivers/scsi/scsi_debug.c +3
-4@@ -3651,7 +3651,7 @@ static int do_device_access(struct sdeb_store_info *sip, struct scsi_cmnd *scp,enum dma_data_direction dir;struct scsi_data_buffer *sdb = &scp->sdb;u8 *fsp; -int i; +int i, total =0 ;/* * Even though reads are inherently atomic (in this driver), we expect @@ -3688,18 +3688,16 @@ static int do_device_access(struct sdeb_store_info *sip, struct scsi_cmnd *scp, fsp + (block * sdebug_sector_size), sdebug_sector_size, sg_skip, do_write);sdeb_data_sector_unlock (sip, do_write); -if (ret != sdebug_sector_size) { - ret += (i * sdebug_sector_size); + total += ret; +if (ret != sdebug_sector_size)break ; - } sg_skip += sdebug_sector_size;if (++block >= sdebug_store_sectors)
drivers/scsi/scsi_debug.c static int do_device_access(struct sdeb_store_info *sip, struct scsi_cmnd *scp, · line 3703
Quoted 10 lines
drivers/scsi/scsi_debug.c +1
-2block =0 ; } - ret = num * sdebug_sector_size;sdeb_data_unlock (sip, atomic); -return ret; +return total; }/* Returns number of bytes copied or -1 if error. */
Thank you for fixing this. It looks good to me.
Colin
Martin K. Petersen
AppliedQuoted 8 lines · John Garry
If the
sg_copy_buffer()call returns less thansdebug_sector_size, then we drop out of the copy loop. However, we still report that we copied the full expected amount, which is not proper.Fix by keeping a running total and return that value.
[...]
Applied to 6.12/scsi-fixes, thanks!
[1/1] scsi: scsi_debug: Fix do_device_access() handling of unexpected SG copy length
https://git.kernel.org/mkp/scsi/c/d28d17a84560
Signature · 1 line
Martin K. Petersen Oracle Linux Engineering